There is a company that has led the domestic steel industry for more than half a century. This company, which has a unique position in the field of cold-rolled special steel, supplies core materials to major domestic automakers with its technology that is specialized in supplying materials for automotive parts.
Recently, it has invested more than 120 billion won to enter the secondary battery material market in earnest, preparing for a second leap forward.
However, the company faced a challenge in the face of a new challenge. It was to predict accurate demand and maintain optimal inventory in a rapidly changing market environment.
In particular, the need to efficiently manage the supply chain, which has become more complex due to the entry of new businesses, has emerged.
To solve these challenges, the company decided to introduce an AI-based demand forecasting system. A traditional manufacturing company has begun a new challenge of digital transformation.
So how did the AI demand forecasting system solve the chronic problems of the manufacturing site? And what changes did the site's practitioners experience?
This article contains vivid examples of how AI technology solved the demand forecasting and inventory management problems that the manufacturing industry is struggling with. In particular, it will provide a concrete blueprint for digital transformation to manufacturing practitioners who are considering smart factories.
Let's take a closer look at the changes in the field brought about by the AI demand forecasting system.
At first, our company was also sceptical about the systematisation of demand forecasting. We thought that it was practically impossible to predict the ever-changing market conditions.
However, we discovered new possibilities in a meeting with CEO Jung Doo-hee. We listened to him explain the modelling method that takes into account external factors while using our company's data as a basis.
In particular, during the meeting with Director Choi Hye-bong, where we explained our company's work processes in detail and discussed forecasting methods, we were convinced that AI solutions could actually be helpful. The method of calculating external environment data indicators separately and reflecting them in the modelling was evaluated as a system optimised for the field.
As new businesses, such as the recent entry into the nickel-plated steel sheet business, have expanded, the accuracy of production volume and raw material demand forecasts has become more important. In particular, the need for an integrated demand forecasting system that encompasses the existing cold-rolled special steel business and new businesses has emerged.
Currently, we manage inventory based on a standard of 1.5 to 2.5 months, but we aim to reduce this to 1 to 2 months through AI solutions. Through this, the company expects to be able to manage its inventory assets with greater flexibility.
In the past, when the optimal order quantity was entered into the system, the previous month's data was simply copied and adjustments were made based on only the general internal situation.
The order quantity was adjusted by the person in charge of ordering based on the information obtained by the salesperson when visiting the company. However, now, decisions can be made based on the scientific predictive data provided by the AI solution.
The biggest challenge in the manufacturing process is managing the detailed data itself. Due to the large amount and volume of data, it was not easy to systematically manage the data, especially since the sizes (width, height, length, etc.) of the specifications of the suppliers were all different.
In particular, demand forecasting related to the automotive industry was the most difficult. In the global market, it is easier to predict because the volume of orders is constant, but in the domestic market, it was not easy to keep up with the orders because Hyundai and Kia frequently place orders on a weekly basis.
In the event of a sudden vehicle discontinuation or export, there was a problem of remaining in long-term inventory if the vehicle was discontinued before ordering. In such cases, there was a situation where the vehicle had to be sold at a low unit price.
At the end of the year, the grade conversion work for long-term inventory is carried out, but since the vehicle is sold at a lower unit price than Class A, there is a loss of cost.
However, these problems have been largely resolved since the introduction of the AI solution. AI has enabled us to proactively identify demand through the predicted values it provides, which has greatly increased the efficiency of inventory management.
In addition, by using the average, upper, and lower limits provided by the AI system as guidelines, we have become more flexible in responding to rapidly changing market conditions.
This is leading to an overall operational efficiency improvement that goes beyond simple inventory reduction. In addition, the accuracy of forecasts is continuously being improved by reflecting information collected from the sales field in the system, which is expected to enable more sophisticated demand forecasting in the long term.
Currently, predictions are only made for some of the items, but we plan to expand this to all items. We also plan to continuously study external environmental factors and incorporate them into the model.
Although there may be cases where the error increases, we plan to continue operating the system and develop it in a way that provides guidelines for decision-making to those in charge.
In addition to the raw material demand forecasting solution, we would like to introduce a raw material price forecasting solution to help us determine the optimal time and quantity for purchasing raw materials.
To this end, our company plans to continue cooperating with IMPACTIVE AI to exchange data and improve performance.
Through these efforts, our customers will be able to build a more scientific and systematic production management system, which will in turn strengthen their competitiveness.
Successful examples of the introduction of AI-based demand forecasting systems show that digital transformation in the manufacturing industry is no longer a story of the distant future. In particular, the tangible results achieved by traditional manufacturing companies through new challenges provide meaningful implications for many companies.
The manufacturing industry is now facing a new turning point. Uncertainty in the supply and demand of raw materials is increasing, market changes are becoming faster, and a new challenge has emerged in the form of ESG management. In this era of change, AI demand forecasting systems will become a necessity, not an option.
What is even more encouraging is that AI systems are evolving beyond simple forecasting to become strategic tools that support corporate decision-making.
Its use is expanding into various areas, including inventory management optimisation, forecasting the time of raw material purchase, and production planning. This will soon become a new competitive edge for the manufacturing industry.
In the midst of the massive digital transformation, manufacturing companies must now overcome their fear of change and begin to take on new challenges. The introduction of the AI demand forecasting system, Deepflow, will be the first step.